Designed, developed and deployed a probabilistic supportability and guidance mechanism for the largest ($500 MM+ revenue) customers of AWS EC2, that provides guidance on pool supportability for instance launches. The 'scaling guidance' mechanism currently in production allows select customers to plan their launches based on relevant trends in EC2 pool health, reducing ad-hoc capacity requests.Designed and developed 'Watchful Eye' - an ML based anomaly detection system that uses an… Show more Designed, developed and deployed a probabilistic supportability and guidance mechanism for the largest ($500 MM+ revenue) customers of AWS EC2, that provides guidance on pool supportability for instance launches. The 'scaling guidance' mechanism currently in production allows select customers to plan their launches based on relevant trends in EC2 pool health, reducing ad-hoc capacity requests.Designed and developed 'Watchful Eye' - an ML based anomaly detection system that uses an ensemble of methods to detect anomalies in the EC2 metric space. Anomalies include off-band customer availability, free and unsellable capacity, suspiciously large lead times .etc. The product has currently onboarded 6 core intersection level metrics and will onboard 15 metrics by end of 2024.Built the EC2 demand to supply matching system that linked critical customer reservations of capacity with incoming hardware in the supply chain to allow for better tracking and expedition of critical hardware.Responsibilities also include reviewing organizational metric definitions - keeping them in line with business objectives and developing measures to quantify supply chain and forecast volatility as EC2 expands its offering portfolio. Show less

Took over the Global Address Management and the Global Address Experience Team comprising close to 15 engineers and 4 engineering managers.Focused primarily on operational stability in the Address Experience space that consisted of Tier-1 services that were part of the checkout process in the Amazon Retail Website.Initiated the architectural overhaul of the Global Address Management system to move from a batch processing framework that moved around large RODB files to a real… Show more Took over the Global Address Management and the Global Address Experience Team comprising close to 15 engineers and 4 engineering managers.Focused primarily on operational stability in the Address Experience space that consisted of Tier-1 services that were part of the checkout process in the Amazon Retail Website.Initiated the architectural overhaul of the Global Address Management system to move from a batch processing framework that moved around large RODB files to a real time system that used a distributed graph database (AWS Neptune).Directly managed other strategic initiatives such as Unified Address Matching (US) and Address Resolution Upgrades (selected geographies world wide) on a day to day basis, enabling the roll out of both under severe resource constraints. Show less

Took over the management of the AWS infrastructure Data Engineering team located in Seattle, unifying Data Sciences and Data Engineering under one umbrella.Lead strategic initiatives to improve data backed coordination between supply chain execution planning and execution in the US, culminating in a Sales and Operation Execution (S&OE) process.Managed the further development of core algorithms and systems - the rack promise system, spares planning (forecasting) and value… Show more Took over the management of the AWS infrastructure Data Engineering team located in Seattle, unifying Data Sciences and Data Engineering under one umbrella.Lead strategic initiatives to improve data backed coordination between supply chain execution planning and execution in the US, culminating in a Sales and Operation Execution (S&OE) process.Managed the further development of core algorithms and systems - the rack promise system, spares planning (forecasting) and value manager.Also lead the data engineering team for the supply chain planning organization. The data engineering team owns and operates the organization wide data lake that is used for archival, analytics and metrics generation. Show less

Promoted to become one of the first principal data scientists in all of Amazon in September 2018. Re-oriented the focus of the team from measurement to sophistication in prediction and greater operational control via data.Developed a machine learning based promise prediction model and that generated probabilistic estimates of rack delivery times, replacing the extant deterministic configuration based ETA system (Promise V1.0).Designed, developed and deployed the Value Manager… Show more Promoted to become one of the first principal data scientists in all of Amazon in September 2018. Re-oriented the focus of the team from measurement to sophistication in prediction and greater operational control via data.Developed a machine learning based promise prediction model and that generated probabilistic estimates of rack delivery times, replacing the extant deterministic configuration based ETA system (Promise V1.0).Designed, developed and deployed the Value Manager system, that controls fulfillment and replenishment policies by imputing a specific value to each fulfillment decision for racks in the data center. The system evolved to use 12 different fulfillment attributes to decide the suitability of making a specific decision with regard to rack placement in data centers and is currently the primary decision making framework for rack fulfillment in the AWS supply chain. Show less

Built a data sciences team (4 Data Scientists) and the reporting infrastructure for analyzing supply chain operations from scratch.Normalized all extant data sources and created relevant data dictionaries and the reference manuals for organization wide use. This was the basis for an organizational level measurement and performance review overhaul.Created component level tracking metrics for the data center supply chain that helped identify component cross-usage and seepage from… Show more Built a data sciences team (4 Data Scientists) and the reporting infrastructure for analyzing supply chain operations from scratch.Normalized all extant data sources and created relevant data dictionaries and the reference manuals for organization wide use. This was the basis for an organizational level measurement and performance review overhaul.Created component level tracking metrics for the data center supply chain that helped identify component cross-usage and seepage from spares by tracking the flow of over 60 Million small and medium sized componentsCreated a cost model to determine the revenue losses from delayed rack delivery delivery and infrastructure hand off that helped identify a $40 Million opportunity for savings annuallyDeveloped the AWS infrastructure scoreboard for display of metrics related to the supply chain at multiple levels relevant to both executives and line managers. Show less

Led the Last Mile Analytics (LMA) organization - a team of 15 business analysts and business intelligence engineers which was responsible for metric development, measurement and data-driven problem solving in Amazon's world wide last mile network.Conceptualized and implemented the Zone Efficiency and Route Optimization (ZERO) algorithm that implemented a preferential density based cost engine for optimizing AMZL deliveries in US and UK. The ZERO project unlocked savings worth $75MM… Show more Led the Last Mile Analytics (LMA) organization - a team of 15 business analysts and business intelligence engineers which was responsible for metric development, measurement and data-driven problem solving in Amazon's world wide last mile network.Conceptualized and implemented the Zone Efficiency and Route Optimization (ZERO) algorithm that implemented a preferential density based cost engine for optimizing AMZL deliveries in US and UK. The ZERO project unlocked savings worth $75MM annually and is currently pending US Patent as a 'novel method to densify and optimize package delivery based on address and area level attributes'Was responsible for standardizing, automating the publishing of all last mile related metrics and orienting them to focus on customer experience related issues rather than a pure operational focus. Show less

Awarded the Free Cash Flow champion award for efforts in reducing delivery defects related to location finding issues by devising the scatter diagonal method based on historical delivery dataLed a team of 4 program managers in developing delivery station specific scorecards for performance management and measurement.